Re: scope of solution..

[Brian E Carpenter <brian@hursley.ibm.com>]

Getting back to the requirements discussion, I repeat myself:

> ... Assuming we currently have the
> BGP table growing like k+N (where N is the number of multihomed sites),
> do we aim at it growing like k+log(N), or what?...

The requirement is clearly to get away from punching a billion holes
in BGP aggregates, and if we achieve that, the solution for passing
hints to the hosts will probably drop right out.

But maybe we should add a conditional requirement

  If the solution relies on use of multiple addresses per host, it must
  include a simple, scaleable, secure mechanism for the routing
  system to tell each host which addresses to use.

      Brian

Bill Sommerfeld wrote:
> 
> > At billions of prefixes, the routers will not be able to do the lookup in
> > a reasonable time anyway so your packets will get there sometime in the
> > next century.
> 
> A host-based multi-address multihoming scheme might not require any
> individual router to support a billion prefixes.
> 
>                                         - Bill